Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Pushing Mixture of Experts to the Limit: Extremely Parameter Efficient MoE for Instruction Tuning

About

The Mixture of Experts (MoE) is a widely known neural architecture where an ensemble of specialized sub-models optimizes overall performance with a constant computational cost. However, conventional MoEs pose challenges at scale due to the need to store all experts in memory. In this paper, we push MoE to the limit. We propose extremely parameter-efficient MoE by uniquely combining MoE architecture with lightweight experts.Our MoE architecture outperforms standard parameter-efficient fine-tuning (PEFT) methods and is on par with full fine-tuning by only updating the lightweight experts -- less than 1% of an 11B parameters model. Furthermore, our method generalizes to unseen tasks as it does not depend on any prior task knowledge. Our research underscores the versatility of the mixture of experts architecture, showcasing its ability to deliver robust performance even when subjected to rigorous parameter constraints. Our code used in all the experiments is publicly available here: https://github.com/for-ai/parameter-efficient-moe.

Ted Zadouri, Ahmet \"Ust\"un, Arash Ahmadian, Beyza Ermi\c{s}, Acyr Locatelli, Sara Hooker• 2023

Related benchmarks

TaskDatasetResultRank
Code GenerationHumanEval
Pass@124.83
1036
Multi-task Language UnderstandingMMLU
Accuracy73.2
876
ReasoningBBH
Accuracy35.4
672
Image ClassificationEuroSAT
Accuracy98.63
569
Code GenerationHumanEval (test)
Pass@143.78
506
Image ClassificationSUN397
Accuracy52.55
441
ClassificationCars
Accuracy50.83
395
Image ClassificationRESISC45
Accuracy92.58
349
Image ClassificationiNaturalist 2018
Top-1 Accuracy78
291
Commonsense ReasoningCommonsense Reasoning (BoolQ, PIQA, SIQA, HellaS., WinoG., ARC-e, ARC-c, OBQA) (test)
BoolQ Accuracy73.15
202
Showing 10 of 30 rows

Other info

Follow for update